What is the origin of auxiliary verbs?

When and why did we start using auxiliary verbs, particularly "do", to ask questions and make negatives?

The rise of 'do' in the history of English

The history of do has long been of interest to historical linguists, and there is an extensive literature on the rise of do in the history of English. The change took place over the course of the Middle English period, with the very earliest uses appearing in the beginning of the 15th century. The change took place over the course of about 300 years, with an inflection point at about 1550. The following chart, adapted from Ellegård (1953) shows the appearance of do in a variety of syntactic contexts.

Explaining why the change occurred requires some basic facts about differences between Middle English and Modern English syntax.

English clausal syntax

Let's first start with the structure of the Modern English clause. Consider the embedded (subordinate) clauses in (1) and (2):

1. I think that John will enter the race.
2. I asked if John will enter the race.

In addition to the the subject John and the main predicate enter the race we also see two other parts of the clause: the auxiliary verb will and the complementizer (or subordinating conjunction) that in (1) and if in (2).

Schematically then, we can think of an English clause as having the following structure:

Complementizer Subject Aux VP

Now in clauses without an auxiliary verb, the verbal inflection shows up as a morphological change (usually) on the main verb. So if we change 'will' to the past tense in (1) and (2) we get (3) and (4).

3. I think that John entered the race.
4. I asked if John entered the race.

Despite the missing Aux, however, we still think there is an Aux, but it is realized as the past tense morphology (here, the suffix '–ed') on the main verb. There are a number of arguments for this analysis, but the simplest one is that if you want to make a sentence with an inflected main verb emphatic, you use do instead of inflecting the verb. (E.g. John DID enter the race. as a response to someone having claimed otherwise.)

Combining these two analyses together, we make the basic structure of the clause remain the same, but instead of saying that every clause has an Aux as in (4) we say that every clause has a Tensed element, and some tenses are free standing words (like the Modal verbs) and some are affixes (like the PAST and PRESENT tense morphology on main verbs.) [Note: I'm abstracting away from tense differences in the Modal verbs (will vs. would) and treating them each as different kinds of Tensed Auxiliary verbs.]

So our new schematic clause is:

Complementizer Subject Tense VP (shorthand version: C Subj T VP)

Complementizers and Questions

Now lets turn to the complementizer. In the examples I've shown, there is a declarative complementizer that and an interrogative complementizer if. Now the main clause version of the declarative clause is simply the clause without the Complementizer (i.e., John entered the race.) But if we think about what the main clause version of if John entered the race is, the complementizer disappears (just like that disappears) but the Auxiliary verb ends up to the left of the subject:

5. Will John enter the race?
6. Did John enter the race?

This shows that there's a close connection between the complementizer position and the placement of the Auxiliary verb. We can therefore say that the structure of (5) and (6) is the same as the corresponding embedded clause with if except that the Auxiliary verb is in the Complementizer position. We'll call this rule "T-to-C": the element in the T position moves to the C position.

T-to-C Rule: Move the word in T to C (obligatory in main clause questions)

In English main clause questions, this rule is obligatory for all regular questions except those involving the subject of the main clause (Who entered the race? not *Who did enter the race?).

The syntax of Negation

Now let's talk about not in Modern English. Descriptively we can see that not always appears to the left of the main verb and immediately after the first Auxiliary in the clause. If there is no Aux, then we must have one, and do is inserted. This pattern in shown in (7)-(10).

7. John will not enter the race.
8. *John not will enter the race.
9. *John not entered the race.
10. John did not enter the race.

This means that our schematic clause now needs a place for not (which we will call Negation) and looks like this:

C Subj T Neg VP

Furthermore, we can now motivate a descriptive rule about not in English: not always requires a free-standing word in T.

NOT-rule: not requires a word in T.

Accounting for the Middle English patterns

We now have the pieces we need to explain the change, and the rise of do.

To do this we need to look at Middle English questions and Middle English negated sentences. Here are some examples (from Kroch 1989):

11. How great and greuous tribulations suffered the Holy Appostyls...?

12. and in thy name have we not cast oute devyls...?

13. ...spoile him of his riches by sondrie fraudes, whiche he perceiueth not.

14. Go, say to hym we wyll not grefe [grieve].

Examples (11) and (13) have no Auxiliary verb, while examples (12) and (14) do. Abstracting away from other differences, we can see that the sentences with the Auxiliary verb show the same basic syntax as modern English, but the ones without show something different: in both (11) and (13) we find the main verb in a position that is no longer possible in Modern English. In (11) (a question), the main verb 'suffered' shows up to the left of the subject, and in (13) the main verb 'perceiueth' (in the which clause) shows up to the left of 'not'.

How can we explain these two facts? Well, recall that we have two rules one for questions and one for not, repeated here:

T-to-C Rule: Move the word in T to C

NOT-rule: not requires a word in T

Both of these rules clearly applied in Middle English too, as we can see from the identical syntax of examples (12) and (14). Since these rules applied, then we have a simple explanation for the word order we find in (11) and (13): Middle English had a rule which allowed the main verb to move to the T position in the clause. This correctly puts the main verb to the left of the negation in (13), satisfying the NOT rule, and correctly predicts that the main verb (since it is in T) will move to C in questions, as it does in (11).

Middle English V-to-T rule: If there is no Aux verb, the main verb moves to T

So the rise of do is directly linked to the loss of the V-to-T rule. Once the main verb can no longer move to T, sentences involving not will require a dummy auxiliary verb to satisfy the (still present) NOT-rule. Similarly, questions in English, which require the T-to-C rule, will also require the dummy Aux do to satisfy that rule.

How did we lose the rule?

Answering the question of why the V-to-T rule was lost is still subject to a lot of debate. It seems likely, however, that it is tied to the loss of the verbal inflectional system in Middle English. Early Middle English had a full set of person/number inflections for verbs, and these were gradually lost over time (see Roberts 1985). The timing of the loss doesn't line up directly with the rise of do; so the exact cause is subject to some debate.

References

Ellegård, Alvar. 1953. The Auxiliary do: The Establishment and Regulation of Its Use in English. Edited by Frank Behre. Gothenburg Studies in English. Stockholm: Almqvist and Wiksell.

Kroch, Anthony. 1989. Reflexes of grammar in patterns of language change. Language Variation and Change 1:199-244.

Roberts, Ian. 1985. Agreement parameters and the development of the English modal auxiliaries. Natural Language and Linguistic Theory 3:21-58.

It might interest some people that the closely related Danish language did not handle the situation the same way, even if Danish underwent the same loss of person/number conjugation; in fact, even more completely than English. Today, no Danish verb is ever conjugated in person or number. Compare:

English: John talks. - Does John talk? - John does not talk.

Danish: John taler. - Taler John? - John taler ikke.

We see that Danish kept the old word order even if going through an even more radical simplification of conjugation. Now, I'm not saying that this rules out the explanation, but it does show that the introduction of auxiliaries is not necessary in the situation. Of course, different languages often handle different situations differently.

I once read somewhere that the introduction of the do-constructions could have been influenced by Celtic languages. Is that ruled out as an explanation now? Perhaps both reasons could have worked together to put a higher pressure on English for the change than on Danish.